Croatia

Multi-task language models show outstanding performance for various natural language understanding tasks with only a single model.
However, these language models inevitably utilize an unnecessarily large number of model parameters, even when used only for a specific task.
In this paper, we propose a novel training-free compression method for multi-task language models using pruning method.
Specifically, we use an attribution method to determine which neurons are essential for performing a specific task.
We task-specifically prune unimportant neurons and leave only task-specific parameters.
Furthermore, we extend our method to be applicable in both low-resource and unsupervised settings. 
Since our compression method is training-free, it uses little computing resources and does not update the pre-trained parameters of language models, reducing storage space usage.
Experimental results on the six widely-used datasets show that our proposed pruning method significantly outperforms baseline pruning methods.
In addition, we demonstrate that our method preserves performance even in an unseen domain setting.

EACL 2023

Task-specific Compression for Multi-task Language Models using Attribution-based Pruning

task-specific compression

skill neuron

language model compression

findings / work in progress

### Welcome to the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 
Continuing its mission of expanding and involving the science community of all European countries, EACL had selected the Ukrainian community for the 16th EACL, which took place online due to the COVID pandemic. Unfortunately, the ongoing war made the organisation in Kyiv impossible. Considering the importance of physical interaction among researchers, especially after the restrictions imposed by the COVID pandemic, in addition to an online mode, the conference will be held in Dubrovnik, Croatia, from 2 to 6 of May, 2023. The original aim of strengthening the connection with the Ukrainian community will not change as our program will feature a dedicated session and a workshop to highlight work on Ukrainian language technologies. As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

You need to register for the conference in order to access this site. Please visit https://2023.eacl.org/registration for more information.

As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

In this paper, we cast Legal Judgment Prediction on European Court of Human Rights cases into an article-aware classification task, where the case outcome is classified from a combined input of case facts and convention articles. This configuration facilitates the model learning some legal reasoning ability in mapping article text to specific case fact text. It also provides an opportunity to evaluate the model's ability to generalize to zero-shot settings when asked to classify the case outcome with respect to articles not seen during training. We devise zero-shot experiments and apply domain adaptation methods based on domain discrimination and Wasserstein distance. Our results demonstrate that the article-aware architecture outperforms straightforward fact classification. We also find that domain adaptation methods improve zero-shot transfer performance, with article relatedness and encoder pre-training influencing the effect.

Zero-shot Transfer of Article-aware Legal Outcome Classification for European Court of Human Rights Cases

Recently, we can obtain a practical abstractive document summarization model by fine-tuning a pre-trained language model (PLM). Since the pre-training for PLMs does not consider summarization-specific information such as the target summary length, there is a gap between the pre-training and fine-tuning for PLMs in summarization tasks.
To fill the gap, we propose a method for enabling the model to understand the summarization-specific information by predicting the summary length in the encoder and generating a summary of the predicted length in the decoder in fine-tuning.
Experimental results on the WikiHow, NYT, and CNN/DM datasets showed that our methods improve ROUGE scores from BART by generating summaries of appropriate lengths. Further, we observed about 3.0, 1,5, and 3.1 point improvements for ROUGE-1, -2, and -L, respectively, from GSum on the WikiHow dataset. Human evaluation results also showed that our methods improve the informativeness and conciseness of summaries.

Abstractive Document Summarization with Summary-length Prediction

Hierarchical text classification (HTC) aims to assign the most relevant labels with the hierarchical structure to an input text. However, handling unseen labels with considering a label hierarchy is still an open problem for real-world applications because traditional HTC models employ a pre-defined label set. To deal with this problem, we propose a generation-based classifier that leverages a Seq2Seq framework to capture a label hierarchy and unseen labels explicitly. Because of no available social media datasets that target at HTC, we constructed a new (Blog) dataset using pairs of social media posts and their hierarchical topic labels. Experimental results on the Blog dataset showed the effectiveness of our generation-based classifier over state-of-the-art baseline models. Human evaluation results showed that the quality of generated unseen labels outperforms even the gold labels.

Hierarchical Label Generation for Text Classification

Current multilingual semantic parsing (MSP) datasets are almost all collected by translating the utterances in the existing datasets from the resource-rich language to the target language. However, manual translation is costly. To reduce the translation effort, this paper proposes the first active learning procedure for MSP (AL-MSP). AL-MSP selects only a subset from the existing datasets to be translated. We also propose a novel selection method that prioritizes the examples diversifying the logical form structures with more lexical choices, and a novel hyperparameter tuning method that needs no extra annotation cost. Our experiments show that AL-MSP significantly reduces translation costs with ideal selection methods. Our selection method with proper hyperparameters yields better parsing performance than the other baselines on two multilingual datasets.

Active Learning for Multilingual Semantic Parser

Language documentation often requires segmenting transcriptions of utterances collected on the field into words and morphemes. While these two tasks are typically performed in succession, we study here Bayesian models for simultaneously segmenting utterances at these two levels. Our aim is twofold: (a) to study the effect of explicitly introducing a hierarchy of units in joint segmentation models; (b) to further assess whether these two levels can be better identified through weak supervision. For this, we first consider a deterministic coupling between independent models; then design and evaluate hierarchical Bayesian models. Experiments with two under-resourced languages (Japhug and Tsez) allow us to better understand the value of various types of weak supervision. In our analysis, we use these results to revisit the distributional hypotheses behind Bayesian segmentation models and evaluate their validity for language documentation data.

Joint Word and Morpheme Segmentation with Bayesian Non-Parametric Models

When humans read a text, their eye movements are influenced by the structural complexity of the input sentences. This cognitive phenomenon holds across languages and recent studies indicate that multilingual language models utilize structural similarities between languages to facilitate cross-lingual transfer. We use sentence-level eye-tracking patterns as a cognitive indicator for structural complexity and show that the multilingual model XLM-RoBERTa can successfully predict varied patterns for 13 typologically diverse languages, despite being fine-tuned only on English data. We quantify the sensitivity of the model to structural complexity and distinguish a range of complexity characteristics. Our results indicate that the model develops a meaningful bias towards sentence length but also integrates cross-lingual differences. We conduct a control experiment with randomized word order and find that the model seems to additionally capture more complex structural information.

Cross-Lingual Transfer of Cognitive Processing Complexity

Script diversity presents a challenge to Multilingual Language Models (MLLM) by reducing lexical overlap among closely related languages. Therefore, transliterating closely related languages that use different writing scripts to a common script may improve the downstream task performance of MLLMs. 
We empirically measure the effect of transliteration on MLLMs in this context. 
We specifically focus on the Indic languages, which have the highest script diversity in the world, and we evaluate our models on the IndicGLUE benchmark. We perform the Mann-Whitney U test to rigorously verify whether the effect of transliteration is significant or not. We find that transliteration benefits the low-resource languages without negatively affecting the comparatively high-resource languages. We also measure the cross-lingual representation similarity of the models using centered kernel alignment on parallel sentences from the FLORES-101 dataset. We find that for parallel sentences across different languages, the transliteration-based model learns sentence representations that are more similar.

Does Transliteration Help Multilingual Language Modeling?

In this paper, we focus on the topics of misinformation and racial hoaxes from a perspective derived from both social psychology and computational linguistics. In particular, we consider the specific case of anti-immigrant feeling as a first case study for addressing racial stereotypes. We describe the first corpus-based study for multilingual racial stereotype identification in social media conversational threads. Our contributions are: (i) a multilingual corpus of racial hoaxes, (ii) a set of common guidelines for the annotation of racial stereotypes in social media texts, and a multi-layered, fine-grained scheme, psychologically grounded on the work by Fiske, including not only stereotype presence, but also contextuality, implicitness, and forms of discredit, (iii) a multilingual dataset in Italian, Spanish, and French annotated following the aforementioned guidelines, and cross-lingual comparative analyses taking into account racial hoaxes and stereotypes in online discussions. The analysis and results show the usefulness of our methodology and resources, shedding light on how racial hoaxes are spread, and enable the identification of negative stereotypes that reinforce them.

A Multilingual Dataset of Racial Stereotypes in Social Media Conversational Threads

Quotes are critical for establishing credibility in news articles. A direct quote enclosed in quotation marks has a strong visual appeal and is a sign of a reliable citation. Unfortunately, this journalistic practice is not strictly followed, and a quote in the headline is often "contextomized." Such a quote uses words out of context in a way that alters the speaker's intention so that there is no semantically matching quote in the body text. We present QuoteCSE, a contrastive learning framework that represents the embedding of news quotes based on domain-driven positive and negative samples to identify such an editorial strategy. The dataset and code are available at https://github.com/ssu-humane/contextomized-quote-contrastive.

Detecting Contextomized Quotes in News Headlines by Contrastive Learning

What are the events involved in a pandemic outbreak? What steps should be taken when planning a wedding? The answers to these questions can be found by collecting many documents on the complex event of interest, extracting relevant information, and analyzing it. We present a new approach in which large language models are utilized to generate source documents that allow predicting, given a high-level event definition, the specific events, arguments, and relations between them to construct a schema that describes the complex event in its entirety.
Using our model, complete schemas on any topic can be generated on-the-fly without any manual data collection, i.e., in a zero-shot manner. Moreover, we develop efficient methods to extract pertinent information from texts and demonstrate in a series of experiments that these schemas are considered to be more complete than human-curated ones in the majority of examined scenarios. Finally, we show that this framework is comparable in performance with previous supervised schema induction methods that rely on collecting real texts and even reaching the best score in the prediction task.

Downloads

Next from EACL 2023

Zero-shot Transfer of Article-aware Legal Outcome Classification for European Court of Human Rights Cases